1. Field of the Invention
The present invention relates to an unevenness correction data production apparatus and an image forming apparatus and, more particularly, is suitable for an image forming apparatus for performing image formation using a recording head having a plurality of recording elements constituting an array.
Particularly, the present invention relates to an apparatus having a mechanism for automatically controlling print characteristics of a recording head in an ink-jet recording apparatus and is especially effective in an apparatus for forming a multi-gradation color image by overlapping ink droplets.
2. Related Background Art
Along with development of information processing equipment (e.g., a copying machine, a wordprocessor, and a computer) and communication equipment, apparatuses for performing digital image recording using a recording head by using an ink-jet scheme, a thermal transfer scheme, or the like have been very popular as image forming (recording) apparatuses for equipment such as information processing equipment and communication equipment. In such a recording apparatus, a recording head having an array of a plurality of recording elements (to be referred to as a multi-head hereinafter) is generally used to increase a recording speed.
For example, a multi-nozzle head having a plurality of ink orifices and a plurality of liquid paths is generally used in an ink-jet recording head. A plurality of heaters are generally stacked in a thermal transfer or thermal head.
It is very difficult to uniformly manufacture recording elements of a multi-head due to variations in characteristics during the fabrication process and variations in properties of head constituting materials. Variations occur in the characteristics of the respective recording elements. For example, in the above multi-nozzle head, variations occur in shapes of the orifices and liquid paths. In the above thermal head, shapes of heaters and resistances inevitably vary. Nonuniformity of characteristics between the recording elements appears as nonuniformity of sizes of dots recorded by the respective recording elements and as uneven image densities of the resultant image.
To cope with the above problems, various methods are proposed in which an uneven image density is visually found or an adjusted image is visually checked, a signal applied to each recording element is manually corrected, thereby obtaining a uniform image.
For example, in a multi-head 330 having recording elements 331 arranged as shown in FIG. 1A, assume that input signals to the respective recording elements are uniformed, as shown in FIG. 1B, and that an uneven image density is visually found, as shown in FIG. 1C. In this case, an input signal is corrected, as shown in FIG. 1D. More specifically, a large input signal is supplied to a recording element corresponding to a low image density, and a small input signal is supplied to a recording element corresponding to a high image density, thereby generally performing manual adjustment.
In a recording scheme capable of modulating a dot diameter or dot density, the diameter of a dot to be recorded by each recording element is modulated in accordance with an input to achieve gradation recording. For example, if modulation of a drive voltage applied to each injection energy generating element (e.g., a piezoelectric element or an electricity-heat conversion element) in an ink-jet recording head according to a piezoelectric scheme or a scheme utilizing heat energy, or a drive voltage applied to each heater in a thermal head or a pulse width of the drive voltage in accordance with an input signal is utilized, a dot diameter or a dot density by each recording element can be uniformed, and a density distribution can be uniformed, as shown in FIG. 1E. Alternatively, assume that it is difficult to modulate a drive voltage or pulse width or it is difficult to perform density adjustment in a wide range even if the drive voltage or pulse width is modulated. In this case, if one pixel is constituted by, e.g., a plurality of dots, the number of dots to be recorded in accordance with an input signal is modulated. A larger number of dots are assigned to a portion having a low density, while a smaller number of dots are assigned to a portion having a high density. If one pixel is constituted by one dot, the number of orifice-use times (the number of injection cycles) per pixel is modulated in the ink-jet recording apparatus, thereby changing the dot diameter. Therefore, the density distribution can be uniformed, as shown in FIG. 1E.
Japanese Patent Laid-Open Application No. 57-41965 filed by the present applicant discloses that a color image is automatically read by an optical sensor, and a correction signal is supplied to each ink-jet recording head to form a desired color image. In this prior-art invention, basic automatic adjustment is disclosed, and an important technique is thus disclosed. Various problems may be posed when this prior-art invention is embodied in a variety of practical applications as various apparatuses. However, technical problems of the present invention are not found in this prior-art invention.
Techniques except for a density detection scheme are disclosed in Japanese Laid-Open Patent Application No. 60-206660, U.S. Pat. 4,328,504, and Japanese Patent Laid-Open Application Nos. 50-147241 and 54-27728. A landing position of a liquid droplet is automatically read, and the read position data is corrected to perform landing to an accurate position. Although these schemes are common as automatic adjustment techniques, the technical problems of the present invention are not found in these prior-art inventions.
In order to cope with the above problem, an uneven image density reading unit is arranged in an image forming apparatus, and an uneven image density distribution within the array of recording elements is periodically read to rewrite an unevenness correction data, thus providing an effective countermeasure. According to this technique, even if the uneven image density distribution of the head is changed, the unevenness correction data is rewritten accordingly, thereby always obtaining a uniform image free from unevenness. Such an image recording apparatus is proposed by the present applicant (U.S. Ser. No. 480,041 filed on Feb. 14, 1990 abandoned; and U.S. Ser. No. 516,129 filed on Apr. 27, 1990 abandoned).
FIG. 2 shows an uneven image density reading unit 506 used in the above method. A recording medium 501 has an unevenness measurement test pattern. The reading unit 506 includes a light source 502 for illuminating a surface of the recording medium, a reading sensor 503 for reading light reflected by the surface of the recording medium, and lenses 504 and 505. The reading unit 506 having the above arrangement is scanned to read an unevenness distribution, thereby rewriting unevenness correction data.
FIG. 3 is a view for explaining a reading mode of a test pattern. A reading unit 506 is the one shown in FIG. 2. A recording head 520 has recording elements 521 aligned within a range l in the direction of width (x direction) of a recording medium. A test pattern 524 has a predetermined width W and is recorded upon appropriate driving of the recording elements 521 during relative movement (i.e., conveyance of the recording medium 501 in the y direction) between the recording medium 501 and the recording head 520. This test pattern 524 is read upon scanning of the recording unit 506 in the x direction.
FIG. 4 shows another uneven image density reading unit. The reading unit comprises a line sensor 520 comprising a CCD or the like having read pixels 521, and an unevenness correction test pattern 524 in which a recording element is formed by a width d in the y direction. The density of the test pattern is read by a recording head while the line sensor 520 is scanned in the x direction. Data read by the pixels 521 of the line sensor 520 correspond to density data formed by the respective recording elements of the recording head.
FIG. 5 shows a relationship between scanning positions during scanning of the reading unit in the x direction and test pattern read densities in FIGS. 3 and 4. As is apparent from FIGS. 3 and 4, density data has moderate leading and trailing edges. It is difficult to employ any point corresponding to a recording element located at the end of the recording head. For this reason, a threshold value of several tens of % of a maximum output is preset, and a point P having the threshold value is assumed as a reference point of the position of the recording element of the end portion of the recording head. The read data are caused to correspond to the recording elements on the basis of the reference point, thereby forming correction data.
A print duty of each pattern is not constant and is changed as needed. For example, a test pattern having a print duty of about 50% is generally used. However, when an uneven image density of a high-density portion is to be concentratedly corrected, a test pattern having a print duty of about 75% is preferably used. When an uneven image density of a highlighted portion is to be concentratedly corrected, a test pattern having a print duty of about 30% is preferably used. In order to obtain an average correction effect throughout the entire density range, three density distributions of 30%, 50%, and 75% are preferably measured to form correction data from their average value.
In this manner, when a common threshold value is used at the time of a change in print duty, an end position detected based on the print duties varies since the read density is changed with a change in print duty.
For example, referring to FIG. 5, when the threshold value T is kept unchanged upon acquirement of an unevenness distribution B by a change in print duty of a head having an unevenness distribution A, the detected reference point becomes a position P', thus causing inaccuracy in correspondence between unevenness data and the recording elements. As a result, accurate correction may not be performed.
When an image forming apparatus has recording heads of two or more colors and unevenness correction is to be performed for these heads, the following problem is posed.
For example, when spectral sensitivity of the reading head shown in FIG. 2 was close to a human luminosity factor according to an experiment of the present applicant, read densities of magenta, cyan, and black heads were respectively 1.44, 1.46 and 1.55 under the condition that a read density of a yellow head is set to 1. In this manner, since read densities are different depending on different colors, when the common threshold value is used, end positions detected for different colors are different from each other.
For example, referring to FIG. 5, when the color of a head having the unevenness distribution A is changed to obtain the unevenness distribution B, and the threshold value B is kept unchanged, the detected reference point is the position P', thus causing inaccurate correspondence between the unevenness data and the recording elements. As a result, accurate correction may not be performed.
Since the reading means described above is mounted in a recording apparatus, it must be simple in structure at low cost. For this purpose, read precision is inevitably limited.
Uneven image densities having even a small difference are visually noticed. In order to read and correct these uneven image densities, highly precise reading is required.
Correction quantities by automatic control are obtained as follows.
A test pattern (FIG. 6A) obtained by driving a plurality of recording elements under the same condition, i.e., by the same drive signal (the drive signal is defined as S.sub.0 in this case) is recorded. An optical density of the test pattern is not uniform due to variations caused by fabrication of the respective recording elements and variations caused by deteriorations over time, as shown in FIG. 6B. An uneven image density is caused. This uneven image density is read, partial densities OD.sub.1 to OD.sub.N corresponding to all the recording elements are measured, and an average density as the correction target is obtained as follows: ##EQU1##
This average density need not be obtained by simply averaging the densities of the all the recording elements. For example, quantities of reflected light may be integrated and the integral values are averaged to obtain an average value, or another known method may be used to obtain an average value.
If a relationship between image signal values and an output density of a given element or a given element group is given as shown in FIG. 7, a signal actually supplied to this element or this element group is obtained by correcting the signal S to obtain a correction coefficient .alpha. for obtaining the target density OD. A correction signal .alpha..multidot.S obtained by correcting the signal S to .alpha..times.S=(OD/OD.sub.n).times.S is supplied to the element n or the element group. More specifically, table conversion shown in FIG. 8 is performed for the input image signal in practice. Referring to FIG. 8, a straight line A is a line having a gradient of 1.0. This table is a table for performing no conversion of an input signal and directly outputting the input signal. A straight line B is a straight line having a gradient of .alpha.=OD/OD.sub.n. This table is a table for converting the input signal S into an output signal .alpha..multidot.S. The table representing the straight line B in FIG. 8 is used for the image signal corresponding to the nth recording element, and table conversion having a correction coefficient .alpha..sub.n is performed. Thereafter, when the head is driven, the densities of the portion recorded by the N recording elements are equal to OD. This processing is performed for all the recording elements to correct the uneven image densities, thereby obtaining a uniform image. That is, when data representing a correspondence between a given table conversion coefficient and an image signal of each recording element is obtained, unevenness correction can be performed.
The above correction for density comparison may be performed for each nozzle group (3 to 5 nozzles) to perform correction in accordance with approximation of unevenness correction.
As briefly described above, however, the following problem is posed by a sequence wherein light is incident on the test pattern, as shown in FIG. 6A, the quantity of light reflected by the test pattern is measured, a correction quantity is calculated by the above arithmetic operation method, and each recording element is driven by a corrected drive signal.
In measurement of optical densities of end portions of a test pattern, as indicated by A and B in FIG. 6A, i.e., measurement of the quantity of light reflected by the test pattern upon its radiation with light, values having many errors caused by flare influences are inevitably measured, and recording by the recording elements corresponding to the portions A and B causes uneven image densities.
FIG. 9B shows a typical distribution of the quantity of reflected light measured by an optical sensor upon radiation of a test pattern shown in FIG. 9A. At the time of recording of a test pattern, all the recording elements are driven in accordance with a common drive signal. As described above, quantities of reflected light are not kept constant due to variations in recording elements, resulting from various causes. An uneven image density is present in a recorded image. End portions outside the pattern and a recording medium are left white. The quantity of reflected light is large, and the quantity of light received by an optical sensor is also large. For this reason, portions A' and B' (FIG. 9B) slightly inside the white portions receive flare influences. When light quantity to density conversion is performed by a correction quantity calculation algorithm using the quantities of light received by the sensor, the optical densities have values smaller than actual values. In this case, a correction value for increasing a drive signal for a recording element so as to increase the optical density is undesirably calculated. As a result, an image obtained by correcting the above correction value may have a higher density at both ends thereof.
In order to essentially solve this problem, flare must be eliminated. An optical system which satisfies this requirement becomes expensive.
This arrangement still has the following points to be improved.
Assume a read range of the sensor 503 under the condition that the reading unit 506 shown in FIG. 2 is located at a given scanning position. When this range is inappropriate, the resultant read signal reflects recording characteristics of a large number of recording elements within this range. Uneven image densities represented by fringes having high spatial frequencies cannot be detected. Accurate reading cannot be performed due to an influence of a difference in the number of dots recorded within this range, and a shortage of a light reception quantity.
The above arrangements still have points to be improved.
At the time of reading of uneven image densities, a distance between the uneven image density reading unit and the recording medium on which measurement test patterns are formed must be kept constant. When this distance is changed, a detection result is changed accordingly.
A recording medium such as paper is often curled at a high or low humidity, and a distance between the uneven image density reading unit and the recording medium is often changed.
In this state, the detection result does not necessarily reflect an accurate uneven image density of a recording head. It is therefore difficult to perform accurate uneven image density correction.
On the other hand, when a recording head is located near a read means, the recording head may adversely affect the read means. For example, in an ink-jet recording apparatus, an ink mist is attached from the recording head to a reading sensor of the read means, or reading precision is degraded by a thermal influence.
When the reading head and the read means are arranged in a single apparatus, the recording head may adversely affect the read means. For example, in an ink-jet recording apparatus, an ink mist or water droplet is attached from the recording head to a read means to degrade reading precision. In this case, accurate correction cannot be performed. Recording paper dust and any other dust may be generally attached to the reading unit.
An uneven image density may be defined as an nonuniform density or discontinuity in a change in density. The uneven image density is typically caused not in a narrow range, e.g., in units of pixels, but in a wide range. This uneven image density can be visually observed by an operator. For this reason, an image reading aperture of a sensor for reading this uneven image density can be set to be larger than the size of a dot recorded by each recording element. Since the width of a change in uneven image density is larger than the above aperture, even if correction data are formed on the basis of data obtained by reading an image at a low resolution, unevenness can be eliminated to a considerable extent.
In correction of uneven image densities, however, in order to improve correction precision, correction data are obtained, an image is read using the resultant correction data, and correction data are obtained again, thereby generally performing correction processing a plurality of times. In this case, as shown in FIG. 10, an image area obtained upon first reading in correction processing and an image area obtained upon second reading may be subjected to aberration in positions on the recording sheet. In this case, when a conventional sensor having a low resolution, i.e., a sensor having a large aperture, is used, it is difficult to specify an end of an image area in accordance with a sensor output. Therefore, a sensor output cannot be caused to correspond to each recording element corresponding to an image area. As a result, appropriate correction of each recording element cannot be performed.
In order to solve this problem, a sensor having a high resolution, i.e., having an aperture in units of pixels can be used. However, since this sensor is expensive, the resultant image forming apparatus becomes expensive accordingly.